- Implement -f and -i options
- Note: This raises our max ping data size to 65507 (0xFFFF - sizeof(IPv4Header) - sizeof(ICMPv4Header)) so we match linux's (iputils) ping max size but Windows' max is 65500 and I'm not sure if we should change our code to match Windows or not
svn path=/branches/aicom-network-branch/; revision=45129
- Fix a bug in the packet waiting code
- Display the IP address if getnameinfo fails
- tracert is actually useful now
svn path=/branches/aicom-network-branch/; revision=45094
- Move the TDI stuff from iphlpapi to a shared library called tdilib
[IP]
- Implement tcpip side of IP_HDRINCL
[PSDK, WSHTCPIP, WSHIRDA]
- Fix definition of WSHGetSocketInformation
[PSDK]
- Add AO_OPTION_* defines
[WSHTCPIP]
- Request notifications for bind also
- Implement WSHSetSocketInformation
[MSAFD]
- Implement event notifications (bonus: fixes a memory leak on socket closure due to unfreed helper context)
- Store TdiConnectionHandle the same way we store TdiAddressHandle
- Half-plement WSPSetSockOpt
- Fix WSPGetSockOpt
[TCPIP]
- Handle AO_OPTION_TTL, AO_OPTION_IP_DONTFRAGMENT (not working yet), AO_OPTION_BROADCAST (not working yet), and AO_OPTION_IP_HDRINCL
- Add new members of ADDRESS_FILE for the preceding AO options
[AFD]
- Return the connection handle in the Information of the IOSB (same as we do with the address file handle)
[GENERAL]
- Tracert works now
svn path=/branches/aicom-network-branch/; revision=45093
- We don't need to be locked anymore at that point and it also causes problems with loopback packets
- Use the TTL value stored in the address file for ICMP packets
svn path=/branches/aicom-network-branch/; revision=45016
- We now use the documented members of NDIS_MINIPORT_BLOCK
- It also fixes the issue of determining if the miniport is currently processing a request
- Remove unneeded IRQL raises
- Half-plement MiniSendResourcesAvailable (NdisMSendResourcesAvailable)
- Comment out the code inside ProSendPackets (only used by protocols, not miniports)
svn path=/branches/aicom-network-branch/; revision=45009
- Change some NdisDprAcquireSpinLock/NdisDprReleaseSpinLock calls to NdisAcquireSpinLock/NdisReleaseSpinLock
svn path=/branches/aicom-network-branch/; revision=45007
- Completely rewrite (again) the locking code and use references to ensure that the connection doesn't get freed while completing requests (the most frequent cause of crashes)
- Remove DrainSignals and complete requests inside HandleSignalledConnection instead of doing them in a separate thread (increases speed a lot)
[OSKITTCP]
- Don't clear the socket context in OskitTCPClose because we would end up in HandleSignalledConnection without a connection (which we don't support anymore after eliminating DrainSignals)
- Change the check performed to see if a socket is dying so we support connection dying after calling OskitTCPClose
[AFD]
- Remove leftover ASSERTs which fail after the changes to tcpip (they were wrong in the first place because we call into tcpip at DISPATCH_LEVEL sometimes)
svn path=/branches/aicom-network-branch/; revision=44839
- Bind to the address specified in AddressFile instead of always binding to the outgoing interface's unicast address
- Get the NCE using the local address (if specified) so we don't end up sending from one interface using another interface's address as the source
svn path=/branches/aicom-network-branch/; revision=44810
- Add ROS-specific MSS code
- We don't mirror our route cache with oskittcp so we can't rely on routes to get the MTU
svn path=/branches/aicom-network-branch/; revision=44619
- Implement TDI_QUERY_MAX_DATAGRAM_INFO
[AFD]
- Send TDI_QUERY_MAX_DATAGRAM_INFO to set our send/receive windows properly
- Don't create a send window for a datagram socket because it is never used
- Fixes dropping all packets with length > 16384
svn path=/branches/aicom-network-branch/; revision=44412
- Check that the socket allows connections in AfdDisconnect and AfdStreamSocketConnect
- Allocate the receive window in connectionless sockets when the socket is bound instead of at FCB creation
- Remove a horrible hack in UnlockRequest
- Remove an incorrect call to MmUnmapLockedPages (the pages aren't mapped!)
- Replace the recursive mutex with a KMUTEX for now (eventually I want to move to a fast mutex)
- Fix a leak in LockRequest in failure case
svn path=/branches/aicom-network-branch/; revision=44370
- TSVN choked repeatedly when attempting to merge ~9000 revs into the branch (tried 3 times on 2 different computers)
- If someone wants to delete aicom-network-fixes, they are welcome to
- Lesson learned: Letting a branch get thousands of revs out of date is a horrible idea
svn path=/branches/aicom-network-branch/; revision=44353
- Add a hack which prints an annoying message and grants access when it should not be. Callers/bugs should be fixed and this commit reverted after that.
See issue #4169 for more details.
svn path=/trunk/; revision=44348
- Fail in DosDateTimeToFileTime if prior time conversion was unsuccessful
- Should fix the NetStats installation, bug #4917, part 2
svn path=/trunk/; revision=44345
- Logic change of r37554 is correct, RtlTimeFieldsToTime doesn't normalize invalid input values - remove these codes
- Related to bug #4917, part 1
svn path=/trunk/; revision=44344
- Add a new makefile config to the global configuration called RosBuild
- Use the new makefile config to process and external batch file which make use of rbuild.
- Split and partially rewrite _generate_vcproj. Configurations are now written in support functions.
- Remove the creation of .vcproj.user files. They're rather pointless for our needs
- Start to add support for VS2010 (make msvc10), it's very incomplete at the moment so don't bother trying it.
After this commit, you can now use Visual Studio to build individual reactos components. Run 'make msvc9', open up any vcproj file, select the RosBuild config and hit the build/rebuild button. Working build features are 'build', 'rebuild' and 'clean'
svn path=/trunk/; revision=44343
It won't currently work without RosBE as it relies in the gcc toolset installed along with RosBE to go about its business.
It currently checks for RosBE existence with a rather hackish ^H^H^H^H ... elaborate string parser using the 'reg' command line util output. I'd be happy for anyone to improve on this.
svn path=/trunk/; revision=44342
AbiWord from 2.6.4 to 2.6.8
MinGW from 5.1.4 to 5.1.6
MirandaIM from 0.8.9 to 0.8.10
Opera from 9.6.4 to 10.10 (YES, this works!)
SDL Mixer from 1.2.9 to 1.2.11
SeaMonkey from 1.1.18 to 2.0 (YES, that too!)
UltraVNC from 1.0.6.5 to 1.0.8.0
uTorrent from 1.8.4 to 1.8.5
vlc from 1.0.2 to 1.0.3
svn path=/trunk/; revision=44328
- Implement support for resetting a stream (not yet used)
- Don't translate the error again in QueryMixerInfo as SyncOverlappedDeviceIoControl has already translated the error
- Don't touch the cbSize member when opening a wave device, as the caller might have passed a WAVEFORMAT struct instead a WAVEFORMATEX struct. Fixes winmm_winetest wave / mixer crashs
- Query all available events when an mixer event is fired
svn path=/trunk/; revision=44326
- Always set Mute control on first channel
[PORTCLS]
- Rewrite node property handling. Fixes lots of bugs and fixes all winmm_winetest mixer failures
svn path=/trunk/; revision=44323
